energies

Article

A New Dynamic Injection System of Urea-WaterSolution for a Vehicular Select CatalystReduction System

Long Li, Wei Lin and Youtong Zhang *

School of Mechanical Engineering, Beijing Institute of Technology, Beijing 100081, China;voyage_lee@163.com (L.L.); lin_wei_sky@163.com (W.L.)* Correspondence: youtong@bit.edu.cn; Tel.: +86-10-6891-5013

Academic Editor: Evangelos G. GiakoumisReceived: 28 September 2016; Accepted: 9 December 2016; Published: 23 December 2016

Abstract: Since the Euro-III standard was adopted, the main methods to inhibit NOx productionin diesel engines are exhaust gas recirculation (EGR) and select catalyst reduction (SCR). On thesemethods SCR offers great fuel economy, so it has received wide attention. However, there alsoexists a trade-off law between NOx conversion efficiency and NH3 slip under dynamic conditions.To inhibit NH3 slip with high NOx conversion efficiency, a dynamic control method for a urea watersolution (UWS) injection was investigated. The variation phenomena of SCR conversion efficiencywith respect to the cross-sensitivity characteristics of the NOx sensor to NH3 have been thoroughlyanalyzed. The methodology of “uncertain conversion efficiency curve tangent analysis” has beenapplied to estimate the concentration of the slipped NH3. The correction factor “ϕ” of UWS injectionis obtained by a comparative calculation of the NOx conversion ability and subsequent NH3 slip.It also includes methods of flow compensation and flow reduction. The proposed control methodhas been authenticated under dynamic conditions. In low frequency dynamic experiments, thiscontrol method has accurately justified the NH3 slip process and inhibits the NH3 emission to a lowerlevel thereby improving the conversion efficiency to a value closer to the target value. The results ofEuropean transient cycle (ETC) experiments indicate that NH3 emissions are reduced by 90.8% andthe emission level of NOx is close to the Euro-V standard.

Keywords: select catalyst reduction (SCR); urea water solution (UWS); NH3 slip; dynamic correction

1. Introduction

Direct injection diesel engines are preferred for their superior economy, power and emissions.Due to the high combustion temperature of the diesel engine, the nitrogen in the air is easily oxidizedby oxygen and produces a large amount of NOx which have a significant pollution impact on theenvironment. Therefore, NOx emissions should be controlled. High pressure fuel injection andturbocharging technology have been used to change the ratio of particulate matter (PM) and NOx

by regulating the fuel injection strategy. From the Euro-II to the Euro-III phase, the high injectionpressure (high common rail fuel injection) system has been used to optimize in-cylinder combustionand regulate the ratio of PM and NOx to reach the emission goals. From the Euro-III to the Euro-IVphase, the main problem is how to significantly reduce PM and NOx emissions. There are two methodsat present: one is to use exhaust gas recirculation (EGR) to reduce in-cylinder NOx, and out of thecylinder, with diesel particulate filter(DPF) to filter PM; the other method is to use select catalystreduction (SCR) to eliminate NOx and PM. In small diesel engines, the SCR system is limited by theexhaust gas temperature, so EGR + DPF technology is used as the main method to solve the emissionproblem. The exhaust temperature of medium and heavy diesel engines is high. At high exhaust

Energies 2017, 10, 12; doi:10.3390/en10010012 www.mdpi.com/journal/energies

Energies 2017, 10, 12 2 of 17

temperatures, diesel engines with SCR + high-pressure common rail (HCR) technology are moreeconomical than diesel engines with EGR + DPF technology, so SCR is more widely used in mediumand heavy duty diesel engines. From the Euro-IV to the Euro-V phase, how to further reduce NOx

has become the key problem. The main method is to optimize the SCR system to enhance the NOx

conversion efficiency and reduce the NH3 leakage [1–6].At present, almost 99% of diesel vehicles work under dynamic operating conditions and thus their

NOx emissions are also a dynamic process. Based on this condition, excellent dynamic performanceis an essential characteristic of the SCR system. SCR control strategies mainly focus on optimizingthe urea water solution (UWS) injection rate algorithms and NH3 slip inhibition. However, there is atrade-off law between NOx conversion efficiency and NH3 slip under dynamic conditions. When theactual injection rate is lower than the theoretical one, the NOx can’t be completely reduced. When theactual injection rate is higher than the theoretical value, NH3 can’t be completely oxidized and thusgenerates secondary pollution [7–9]. Furthermore, NH3 storage and catalyst release make the NH3 slipinhibition more difficult.

Some researchers believe that an oxidation processor installed at the end of the exhaust pipe mayinhibit NH3 slip [10]. Nova and Tronconi [11] added an Ammonia Slip Catalyst (ASC) to the exhaustpipe downstream of the SCR system and completed some investigations by experiment and simulation.The results showed that the studied ASC could efficiently clean up the slipped NH3. Shrestha et al. [12]did some experiments and simulation research on multi-functional wash coated monolith catalysts.They compared the catalysts for a range of temperatures, space velocities, and feed compositions inthe presence of H2O and CO2. Based on the data acquired from the experiments, a dynamic model ofthe NH3 oxidation process was established.

Some researchers supposed that it is necessary to investigate the processes of NH3 storage, releaseand reduction reaction as well as the SCR catalyst [13,14]. Rauch et al. [15] monitored the ammonialoading of a vanadia-based SCR catalyst by a microwave-based method. Their experimental resultsshowed that the method can be applied to different temperatures. It was also possible to determinethe storage of ammonia from the ammonia-to-NOx feed ratio. Zhang and Wang [16] focused onthe simultaneous estimation of ammonia coverage ratios and input. They configured a three-statenonlinear model with the high-gain observer method by assuming the states of the SCR system arehomogenous inside and the SCR cell was a continuous stirred tank reactor. The NOx sensor wascross-sensitive to the NH3 concentration, and the NOx sensor reading was corrected by precise NH3

sensor measurements. The simulation results showed that the designed observer worked well.NOx sensors are used to measure the NOx concentration downstream from the SCR system and

feed it back to the SCR controller. However, research results have showed that NOx sensors have anenhanced cross sensitivity to NH3 [17–19]. According to the structural characteristics of NOx sensors,the NH3 inside the sensor is easily oxidized to NOx. There are mainly three chemical reactions in thisprocess, as described in Equations (1)–(3) [20–23]:

2NH3 + 2O2 → N2O + 3H2O (1)

2NH3 + 2.5O2 → 2NO + 3H2O (2)

2NH3 + 3.5O2 → 2NO2 + 3H2O (3)

Wang [24] believed that the main factor is temperature, which may affect the three chemicalreactions. The cross sensitivity factor of the NOx sensor is changed as the temperature changes.Experiments proved that this factor was between 0.5 and 2 for a range of diesel engineexhaust temperatures.

This paper focuses on the trade-off law between NOx conversion efficiency and NH3 slip by usinga presented method of “uncertain conversion efficiency curve tangent analysis” based on the NH3

cross sensitivity characteristics of the NOx sensor. The degree of NH3 slip will be obtained from thecalculation of the parameters which may affect the shape and locations of this tangent. Subsequently,

Energies 2017, 10, 12 3 of 17

the UWS injection correction factor “ϕ” will be calculated with the dynamic flow compensationand flow reduction. Finally the accuracy of the UWS correction model and the effectiveness ofNH3 slip inhibition will be verified under low-frequency and high-frequency (ETC cycle) dynamicworking conditions.

2. Correction Strategy Mathematical Analysis

For the SCR system control strategy, the corrected UWS injection rate is calculated by Equation (4):

quws,Act = (1 +ϕ)× quws,Bas (4)

where qUWS,Act is the real-time UWS injection rate after correction, qUWS,Bas is the basic UWS injectionrate before correction and ϕ is the correction factor of the UWS injection rate. The Simulink model ofthe correction strategy is shown in Figure 1.

Energies 2017, 10, 12 3 of 17

effectiveness of NH3 slip inhibition will be verified under low-frequency and high-frequency (ETC cycle) dynamic working conditions.

2. Correction Strategy Mathematical Analysis

For the SCR system control strategy, the corrected UWS injection rate is calculated by Equation (4):

( )= + ϕ ×uws,Act uws,Bas1q q (4)

where qUWS,Act is the real-time UWS injection rate after correction, qUWS,Bas is the basic UWS injection rate before correction and φ is the correction factor of the UWS injection rate. The Simulink model of the correction strategy is shown in Figure 1.

Figure 1. The correction strategy model. UWS: urea water solution.

The model consists of two sub-models. The “Basic UWS Injection Rate Model” sub-model collects three pieces of data: (1) engine operating data EngineMsg, which include speed, torque, original emissions, etc.; (2) exhaust gas processor data EGPMsg, which include exhaust temperature before and after the catalyst, gas flow, etc.; (3) injection system data UDSMsg. These data are combined to calculate qUWS,Bas.

The “UWS Correction Model” sub-model collects four sets of data: (1) engine operating data; (2) waste gas treatment data; (3) urea injection system data; and (4) NOx sensor data. These data are processed in the module to obtain the UWS injection rate correction factor φ. The qUWS,Act is calculated using the factor φ and qUWS,Bas.

The change rule of qUWS,Act can be obtained from Equation (5):

( )∂ ∂∂ϕ= = + + ϕ∂ ∂ ∂

uws,Act uws,BasUWS,Act uws,Bas 1

q qa q

t t t (5)

where aUWS,Act is the acceleration of qUWS,Act, and t is the time. The “φ” (the initial value is “0”) is the key factor to correct the UWS injection and keep the SCR

system at a low NH3 slip level with a high conversion efficiency. The UWS correction model is shown in Figure 2.

Figure 1. The correction strategy model. UWS: urea water solution.

The model consists of two sub-models. The “Basic UWS Injection Rate Model” sub-model collectsthree pieces of data: (1) engine operating data EngineMsg, which include speed, torque, originalemissions, etc.; (2) exhaust gas processor data EGPMsg, which include exhaust temperature beforeand after the catalyst, gas flow, etc.; (3) injection system data UDSMsg. These data are combined tocalculate qUWS,Bas.

The “UWS Correction Model” sub-model collects four sets of data: (1) engine operating data;(2) waste gas treatment data; (3) urea injection system data; and (4) NOx sensor data. These data areprocessed in the module to obtain the UWS injection rate correction factor ϕ. The qUWS,Act is calculatedusing the factor ϕ and qUWS,Bas.

The change rule of qUWS,Act can be obtained from Equation (5):

aUWS,Act =∂quws,Act

∂t= quws,Bas

∂ϕ

∂t+ (1 +ϕ)

∂quws,Bas

∂t(5)

where aUWS,Act is the acceleration of qUWS,Act, and t is the time.The “ϕ” (the initial value is “0”) is the key factor to correct the UWS injection and keep the SCR

system at a low NH3 slip level with a high conversion efficiency. The UWS correction model is shownin Figure 2.

Energies 2017, 10, 12 4 of 17Energies 2017, 10, 12 4 of 17

Figure 2. The UWS correction model.

The UWS correction model consists of five sub-models:

The “KT Model” sub-model calculates the real-time NH3 sensitivity factor of the sensor based on the exhaust gas temperature near the NOx sensor. The NOx emission was measured by a NOx sensor under dynamic conditions. The surrounding NH3 may be converted into NOx easily in the NOx sensor. Therefore, the values of NH3 and NOx at the same time will be influenced by the data which is measured by the NOx sensor, as given by Equation (6):

= +3N,Act NO ,Act T NH ,Actx

C C K C (6)

where CN,Act is the NOx concentration measured by the NOx sensor. CNOx,Act is the actual NOx concentration at the testing position. KT is the NH3 cross sensitivity factor of the NOx sensor which could be obtained from the KT map and CNH3,Act is the actual NH3 concentration at the testing position.

The “Engine Emission Model” sub-model is used to calculate the original engine NOx emissions, the target conversion efficiency, the ammonia-nitrogen ratio and the target NOx emission concentration which would support service for the other models as shown in Figure 3. For example the targeted conversion efficiency could be calculated using Equation (7):

= ⋅NO ,Trg NO ,Ori Con,Trgx xC C P (7)

where PCon,Trg is the target conversion efficiency (the highest value without NH3 slip) which can be obtained from the engine emission map, CNOx,Ori is the original NOx concentration of the engine before after treatment and can be obtained by inserting the value calculation of the steady map and CNOx,Trg is the target NOx concentration.

The “NH3 Slip Situation Model” sub-model is based on the output of the first two models to determine the current NH3 leak situation; more details can be seen in Section 2.1. The “UWS Flow Dynamic Reduction Model” sub-model is triggered when an NH3 leak occurs. When there is no NH3 leakage, it is necessary to consider whether there is little UWS injection and trigger the “UWS Flow Dynamic Compensation Model”.

The “UWS Flow Dynamic Compensation Model” and “UWS Flow Dynamic Reduction Model” sub-models are used to calculate the correction factor and compensation factor of the UWS injection rate, respectively (see Sections 2.2 and 2.3 for more details).

Figure 2. The UWS correction model.

The UWS correction model consists of five sub-models:

The “KT Model” sub-model calculates the real-time NH3 sensitivity factor of the sensor basedon the exhaust gas temperature near the NOx sensor. The NOx emission was measured by a NOx

sensor under dynamic conditions. The surrounding NH3 may be converted into NOx easily in theNOx sensor. Therefore, the values of NH3 and NOx at the same time will be influenced by the datawhich is measured by the NOx sensor, as given by Equation (6):

CN,Act = CNOx ,Act + KTCNH3,Act (6)

where CN,Act is the NOx concentration measured by the NOx sensor. CNOx,Act is the actual NOx

concentration at the testing position. KT is the NH3 cross sensitivity factor of the NOx sensor whichcould be obtained from the KT map and CNH3,Act is the actual NH3 concentration at the testing position.

The “Engine Emission Model” sub-model is used to calculate the original engine NOx emissions,the target conversion efficiency, the ammonia-nitrogen ratio and the target NOx emission concentrationwhich would support service for the other models as shown in Figure 3. For example the targetedconversion efficiency could be calculated using Equation (7):

CNOx ,Trg = CNOx ,Ori · PCon,Trg (7)

where PCon,Trg is the target conversion efficiency (the highest value without NH3 slip) which can beobtained from the engine emission map, CNOx,Ori is the original NOx concentration of the engine beforeafter treatment and can be obtained by inserting the value calculation of the steady map and CNOx,Trg

is the target NOx concentration.The “NH3 Slip Situation Model” sub-model is based on the output of the first two models to

determine the current NH3 leak situation; more details can be seen in Section 2.1. The “UWS FlowDynamic Reduction Model” sub-model is triggered when an NH3 leak occurs. When there is no NH3

leakage, it is necessary to consider whether there is little UWS injection and trigger the “UWS FlowDynamic Compensation Model”.

Energies 2017, 10, 12 5 of 17

The “UWS Flow Dynamic Compensation Model” and “UWS Flow Dynamic Reduction Model”sub-models are used to calculate the correction factor and compensation factor of the UWS injectionrate, respectively (see Sections 2.2 and 2.3 for more details).Energies 2017, 10, 12 5 of 17

Figure 3. Low frequency dynamic process.

2.1. NH3 Slip Situation Analysis

In order to justify and analyze the real-time NH3 slip situation of the engine, a method called “uncertain conversion efficiency curve tangent analysis” is presented. From the real-time measured value CN,Act, the uncertain conversion efficiency PCon,Fuz can be calculated using Equation (8). The absolute conversion efficiency PCon,Abs can be obtained from the calculated valve CNOx,Act by Equation (9):

−= = −NO ,Ori N,Act N,Act

Con,FuzNO ,Ori NO ,Ori

1x

x x

C C CP

C C (8)

( ) − −− = = = +3 3NO ,Ori N,Act T NH ,ActNO ,Ori NO ,Act T NH ,Act

Con,Abs Con,FuzNO ,Ori NO ,Ori NO ,Ori

xx x

x x x

C C K CC C K CP P

C C C (9)

According to the results of Equations (8) and (9), the relative conversion efficiency can be calculated with Equation (10):

= − = + −3T NH ,ActCon,Rel Con,Abs Con,Trg Con,Fuz Con,Trg

NO ,Orix

K CP P P P P

C (10)

The change rules of PCon,Fuz, PCon,Abs, and PCon,Rel can be obtained from Equations (11)–(13):

•

∂ = = −

∂

N,Act

NO ,OriCon,Fuz Con,Fuz

x

CC

v Pt

(11)

•

∂ + ∂ = =∂

3T NH ,ActCon,Fuz

NO ,OriCon,Abs Con,Abs

x

K CP

Cv P

t

(12)

∂ + ∂ − ∂ = − =

∂

3T NH ,ActCon,Fuz Con,Trg

NO ,OriCon,Rel Con,Abs Con,Trg

x

K CP P

Cv v v

t

(13)

where vCon,Abs, vCon,Rel, and vCon,Fuz are their velocities. There are several kinds of NH3 slip situations, as follows:

(1) PCon,Fuz > PCon,Trg

For the original map of PCon,Trg obtained from the engine calibration experiments, from the theoretically point of view, with the PCon,Fuz ≤ PCon,Trg under any circumstances. In the actual conditions when the engine calibration points are not enough, engine working instability or

Figure 3. Low frequency dynamic process.

2.1. NH3 Slip Situation Analysis

In order to justify and analyze the real-time NH3 slip situation of the engine, a method called“uncertain conversion efficiency curve tangent analysis” is presented. From the real-time measuredvalue CN,Act, the uncertain conversion efficiency PCon,Fuz can be calculated using Equation (8).The absolute conversion efficiency PCon,Abs can be obtained from the calculated valve CNOx,Act byEquation (9):

PCon,Fuz =CNOx ,Ori − CN,Act

CNOx ,Ori= 1− CN,Act

CNOx ,Ori(8)

PCon,Abs =CNOx ,Ori − CNOx ,Act

CNOx ,Ori=

[CNOx ,Ori −

(CN,Act − KTCNH3,Act

)]CNOx ,Ori

= PCon,Fuz +KTCNH3,Act

CNOx ,Ori(9)

According to the results of Equations (8) and (9), the relative conversion efficiency can becalculated with Equation (10):

PCon,Rel = PCon,Abs − PCon,Trg = PCon,Fuz +KTCNH3,Act

CNOx ,Ori− PCon,Trg (10)

The change rules of PCon,Fuz, PCon,Abs, and PCon,Rel can be obtained from Equations (11)–(13):

vCon,Fuz =•

PCon,Fuz = −∂(

CN,ActCNOx ,Ori

)∂t

(11)

vCon,Abs =•

PCon,Abs =∂PCon,Fuz + ∂

(KTCNH3,ActCNOx ,Ori

)∂t

(12)

vCon,Rel = vCon,Abs − vCon,Trg =∂PCon,Fuz + ∂

(KTCNH3,ActCNOx ,Ori

)− ∂PCon,Trg

∂t(13)

where vCon,Abs, vCon,Rel, and vCon,Fuz are their velocities. There are several kinds of NH3 slip situations,as follows:

Energies 2017, 10, 12 6 of 17

(1) PCon,Fuz > PCon,Trg

For the original map of PCon,Trg obtained from the engine calibration experiments, from thetheoretically point of view, with the PCon,Fuz ≤ PCon,Trg under any circumstances. In the actualconditions when the engine calibration points are not enough, engine working instability orsensor testing errors might occur and lead to an abnormal circumstance (like PCon,Fuz > PCon,Trg).For such an instance the NH3 slip is assumed to be zero, thus the UWS need not be corrected.

(2) 0 ≤ PCon,Fuz ≤ PCon,Trg, and tanθ < 0 (vCon,Fuz < 0)

In this case, the uncertain conversion efficiency is lower than its target and stays away from thetarget value gradually. According to Equation (8), PCon,Fuz becomes smaller due to the increase ofthe CN,Act. The enlargement of the CN,Act may be caused by the following two cases:

• The first case is that the excessively injected UWS caused an acceleration of the process andsubsequently an increasing NH3 slip due unreacted ammonia.

• The second case is that insufficient UWS may cause a slowing the process and lead to agrowing amount of NOx remaining unreduced due to unavailability of reactant.

Therefore, it may be concluded that with the condition aUWS,Act ≤ 0 and PCon,Abs ≤ PCon,Trg,there is no NH3 slip, thus UWS compensation could be continued. When aUWS,Act > 0 andPCon,Abs = PCon,Trg, NH3 slip is severely increased, thus UWS injection should be reduced.

(3) 0 ≤ PCon,Fuz ≤ PCon,Trg, and tanθ ≥ 0 (vCon,Fuz ≥ 0)

In this case, the uncertain conversion efficiency is lower than its target and becomes close tothe target value gradually. In this case it can be concluded that when aUWS,Act > 0 and PCon,Abs≤ PCon,Trg, there is no NH3 slip like the previous cases, thus UWS compensation should becontinued. With the condition aUWS,Act ≤ 0 and PCon,Abs ≥ PCon,Trg, UWS injection should bereduced as NH3 slip is going to increase.

(4) PCon,Fuz < 0

This particular case emerges on ruling out the test error and the engine calibration map error,thus under these circumstances CNOx,Act ≤ CNOx,Ori (theoretically), whereas, CN,Act > CNOx,Ori

(PCon,Fuz < 0), CNH3,Act > 0 as shown in Equation (7). This case indicates a seriously high level ofNH3 slip therefore UWS injection must be stopped immediately.

2.2. Urea Water Solution Flow Dynamic Compensation

There was no NH3 slip during the process of the UWS dynamic compensation. Therefore:{CNH3,Act ≡ 0

CNOx ,Act ≡ CN,Act(14)

It is indicated that the actual amount of injected UWS (including the NH3 released from thecatalyst) was less than the demand of SCR reaction. The condition is 0 ≤ PCon,Abs ≤ PCon,Trg and zeroNH3 slip. Therefore, the correction factor ϕ can be calculated using Equation (15):

ϕ ≡•

∆QNOx ,Red•

QNOx ,ActRed

(15)

Energies 2017, 10, 12 7 of 17

where QNOx,Red (NOx conversion potential) is the difference between the target value and actual valueof the total reduced NOx in a period as shown by Equation (16) and QNOx,AcRed is the actual value ofthe total reduced NOx in a specific period given by Equation (17):

∆QNOx ,Re =∫

CNOx ,OriPCon,TrgqExhdt−∫(CNOx ,Ori − CNOx ,Act)qExhdt

=∫ [

CN,Act − CNOx ,Ori(1− PCon,Trg

)]qExhdt

(16)

QNOx ,ActRe =∫

(CNOx ,Ori − CNOx ,Act)qExhdt =∫

(CNOx ,Ori − CNOx ,N)qExhdt (17)

ϕ =∂∆QNOx ,Red/∂t

∂QNOx ,ActRed/∂t=

[CN,Act − CNOx ,Ori

(1− PCon,Trg

)]qExh

(CNOx ,Ori − CN,Act)qExh=

CNOx ,OriPCon,Trg

(CNOx ,Ori − CN,Act)− 1 (18)

CN,Act ≤ CNOx ,Ori (19)

For the control method, the calculation of UWS injection rate and its acceleration may beaccomplished with Equation (20) or Equation (21):

quws,Act =CNOx ,OriPCon,Trg

(CNOx ,Ori−CNOx ,N)quws,Bas

auws,Act = quws,Bas

∂

(CNOx ,OriPCon,TrgCNOx ,Ori−CN,Act

)∂t +(

CNOx ,OriPCon,TrgCNOx ,Ori−CN,Act

)∂quws,Bas

∂t

CN,Act ≤ CNOx ,Ori

(20)

quws,Act = 0auws,Act = 0

CN,Act ≤ CNOx ,Ori

(21)

where NH3 slip is assumed to be zero.The value of NOx emission in this process may be obtained with:{

QNOx =∫

CN,ActqExhdtQNH3 ≡ 0

(22)

2.3. Urea Water Solution Flow Dynamic Reduction

During the UWS dynamic process, reduction is the response of increasing NH3 slip. Therefore:{CNH3,Act 6= 0

CNOx ,Act + KT · CNH3,Act = CN,Act(23)

The case of 0≤ PCon,Abs ≤ PCon,Trg and presence of evident NH3 slip shows that the actual amountof injected UWS (including the NH3 released from the catalyst) is much more than the demand ofthe SCR reaction. This current situation indicates that the SCR reaction is saturated as shown byEquation (24):

CNOx ,Act = CNOx ,Ori(1− PCon,Trg

)(24)

According to Equation (7):

CNH3,Act =CN,Act − CNOx ,Ori

(1− PCon,Trg

)KT

(25)

Energies 2017, 10, 12 8 of 17

The correction factor ϕ can be calculated as Equation (26):

ϕ = −•

QNH3

RAN

•(QNOx ,TrgRed +

QNH3RAN

) = −•

QNH3•(

RANQNOx ,TrgRed + QNH3

) (26)

where QNH3 is the total NH3 emission amount in a specific period of time given by Equation (27),QNOx,TrgRed is the total reduced NOx with target conversion efficiency of Equation (28) and RAN is theammonia nitrogen ratio constant set in the SCR system control strategy:

QNH3 =∫

CNH3,ActqExhdt =∫ [CN,Act − CNOx ,Ori

(1− PCon,Trg

)KT

]qExhdt (27)

ϕ = − ∂QNH3 /∂t

∂(RANQNOx ,TrgRed+QNH3)/∂t

=[CNOx ,Ori(1−PCon,Trg)−CN,Act]qExh

KT

(RANPCon,TrgCNOx ,Ori−

CNOx ,Ori(1−PCon,Trg)−CN,ActKT

)qExh

=CNOx ,Ori(1−PCon,Trg)−CN,Act

KT RANPCon,TrgCNOx ,Ori−CNOx ,Ori(1−PCon,Trg)+CN,Act

(28)

Due to qUWS,Act ≥ 0 and 1 + ϕ ≥ 0:

CN,Act ≥ CNOx ,Ori − (1 + KT RAN)PCon,TrgCNOx ,Ori (29)

In the presence of NH3 Slip, the control method may be applied to calculate the UWS injectionrate and its acceleration is given by Equation (30) or Equation (31):

quws,Act =(

1− 1KT RAN

+CNOx ,Ori−CN,Act

KT RANPCon,TrgCNOx ,Ori

)quws,Bas

auws,Act = quws,Bas

∂

(CNOx ,Ori−CN,Act

RANPCon,TrgCNOx ,Ori− 1

KT RAN

)∂t +(

1− 1KT RAN

+CNOx ,Ori−CN,Act

KT RANPCon,TrgCNOx ,Ori

)∂quws,Bas

∂t

CN,Act ≥ CNOx ,Ori − (1 + KT RAN)PCon,TrgCNOx ,Ori

(30)

quws,Act = 0auws,Act = 0

CN,Act ≥ CNOx ,Ori − (1 + KT RAN)PCon,TrgCNOx ,Ori

(31)

The amount of NOx emission and NH3 emission in this process could be determined by:{QNOx =

∫CN,ActqExhdt =

∫CNOx ,Ori

(1− PCon,Trg

)qExhdt

QNH3 =∫ [

CN,Act − CNOx ,Ori(1− PCon,Trg

)]qExhdt

(32)

2.4. Special Case

For the special case of PCon,Abs > PCon,Trg, the actual value is more than the target value of thesystem conversion efficiency and NH3 slip is assumed to be zero, as can be seen Equation (33):{

CNH3,Act ≡ 0CNOx ,Act ≡ CN,Act

(33)

Energies 2017, 10, 12 9 of 17

The UWS injection rate doesn’t require any dynamic adjustment thus ϕ ≡ 0. Therefore, it can bedescribed with Equation (34) as follows:{

quws,Act = quws,Bas

auws,Act =∂quws,Bas

∂t(34)

The amount of NOx emission and NH3 emission in this process are:{QNOx =

∫CN,ActqExhdt

QNH3 ≡ 0(35)

While in another special case where PCon,Fuz < 0 indicates that NH3 slip is very high. Thus UWSinjection must be stopped immediately. Now ϕ ≡ 0, and:{

quws,Act = 0auws,Act = 0

(36)

The amount of NOx emission and NH3 emission in this case can be described as follows:{QNOx =

∫CN,ActqExhdt =

∫CNOx ,Ori

(1− PCon,Trg

)qExhdt

QNH3 =∫ [

CN,Act − CNOx ,Ori(1− PCon,Trg

)]qExhdt

(37)

3. Experiments and Result Analysis

The low-frequency dynamic working conditions of the engine are reproduced as shown in Figure 1.The detailed dynamic process is shown in Figure 3. The related parameters of the SCR system arechanged in a slower manner for this process by an explicit analysis of their relationships and interactionfactors. Changes of NOx emission and NH3 slip are compared before and after the UWS dynamiccorrection. The ETC cycle was adopted for the high-frequency dynamic process for further verificationof the UWS control method performance. The engine experiment platform is shown in Figure 4.

Energies 2017, 10, 12 9 of 17

2.4. Special Case

For the special case of PCon,Abs > PCon,Trg, the actual value is more than the target value of the system conversion efficiency and NH3 slip is assumed to be zero, as can be seen Equation (33):

≡ ≡

3NH ,Act

NO ,Act N,Act

0

x

C

C C (33)

The UWS injection rate doesn’t require any dynamic adjustment thus ϕ ≡ 0. Therefore, it can be described with Equation (34) as follows:

=

∂= ∂

uws,Act uws,Bas

uws,Basuws,Act

q qq

at

(34)

The amount of NOx emission and NH3 emission in this process are:

=

≡

3

NO N,Act Exh

NH

d

0x

Q C q t

Q (35)

While in another special case where PCon,Fuz < 0 indicates that NH3 slip is very high. Thus UWS injection must be stopped immediately. Now ϕ ≡ 0, and:

= =

uws,Act

uws,Act

00

qa

(36)

The amount of NOx emission and NH3 emission in this case can be described as follows:

( )( )

= = −

= − −

3

NO N,Act Exh NO ,Ori Con,Trg Exh

NH N,Act NO ,Ori Con,Trg Exh

d 1 d

1 dx x

x

Q C q t C P q t

Q C C P q t (37)

3. Experiments and Result Analysis

The low-frequency dynamic working conditions of the engine are reproduced as shown in Figure 1. The detailed dynamic process is shown in Figure 3. The related parameters of the SCR system are changed in a slower manner for this process by an explicit analysis of their relationships and interaction factors. Changes of NOx emission and NH3 slip are compared before and after the UWS dynamic correction. The ETC cycle was adopted for the high-frequency dynamic process for further verification of the UWS control method performance. The engine experiment platform is shown in Figure 4.

Figure 4. Engine experiment platform block diagram. 1: Fuel consumption; 2: Air flowmeter; 3: Dynamometer; 4: Electronic control unit (ECU); 5: Unified diagnostic services (UDS); 6: SCR control

Figure 4. Engine experiment platform block diagram. 1: Fuel consumption; 2: Air flowmeter; 3:Dynamometer; 4: Electronic control unit (ECU); 5: Unified diagnostic services (UDS); 6: SCR controlunit (SCU); 7: UWS tank; 8: Air pump; 9: Monitor; 10: Exterior gateway protocol (EGP); 11: Temperaturesensor; 12: NOx sensor; 13: Emissions analyzer; and 14: Diesel engine.

The specifications of main equipment in the engine experiment platform are listed in Table 1.

Energies 2017, 10, 12 10 of 17

Table 1. Specifications of main equipment.

Name Type Manufacturer Location Remark

Diesel engine ISDe270 40 DCEC Xiangfan, China L6Dynamometer GWD300 POWERLINK Changsha, China Eddy current style

Fuel consumption FC2210 POWERLINK Changsha, China Quality styleAir flow meter ToCeil Shanghai ToCeil Engine Testing Equipment Shanghai, China Hot film style

Emissions analyzer SESAM4.0 AVL Graz, Austria Fourier transforminfrared spectroscopy

The heavy duty diesel engine parameters used in the experiment are given in Table 2.

Table 2. Parameters of ISDe270 40 diesel engine.

CylinderNumber Bore/Stroke Capacity Compression

Ratio Rated Power/Speed Max Torque Fuel Supply Type

- mm/mm L - kW/rpm Nm/rpm -L6 107/124 6.7 17.3:1 198/2500 970/1400 high pressure common rail

3.1. Mathematical Model Validation

As shown in Figure 5 (in this paper, A indicates the values after correction, B indicates the valuesbefore correction, T indicates the values obtained from equipment testing, C indicates the valuesobtained from calculation and M indicates the values obtained from the maps). The UWS injection wasstarted at the 6th second. It can be seen in Figure 5 that the CN,Act curve declined gradually in the first30 s, became stable at a very low level in the second 30 s, and two noticeable humps can be observedin the last 60 s. Considering the NH3 cross sensitivity of the NOx sensor, it could be initially assumedthat the conversion efficiency was low in the first 30 s as the UWS injection was not sufficient. TheNH3 slip increased significantly in the position of the two humps with the severely overloaded UWSinjection. The UWS correction under dynamic conditions is critical for improving SCR conversionefficiency and NH3 slip inhibition.

Energies 2017, 10, 12 10 of 17

unit (SCU); 7: UWS tank; 8: Air pump; 9: Monitor; 10: Exterior gateway protocol (EGP); 11: Temperature sensor; 12: NOx sensor; 13: Emissions analyzer; and 14: Diesel engine.

The specifications of main equipment in the engine experiment platform are listed in Table 1.

Table 1. Specifications of main equipment.

Name Type Manufacturer Location Remark Diesel engine ISDe270 40 DCEC Xiangfan, China L6 Dynamometer GWD300 POWERLINK Changsha, China Eddy current style

Fuel consumption FC2210 POWERLINK Changsha, China Quality style

Air flow meter ToCeil Shanghai ToCeil Engine Testing

Equipment Shanghai, China Hot film style

Emissions analyzer SESAM4.0 AVL Graz, Austria Fourier transform infrared

spectroscopy

The heavy duty diesel engine parameters used in the experiment are given in Table 2.

Table 2. Parameters of ISDe270 40 diesel engine.

Cylinder Number

Bore/Stroke Capacity Compression

Ratio Rated Power/Speed Max Torque Fuel Supply Type

- mm/mm L - kW/rpm Nm/rpm - L6 107/124 6.7 17.3:1 198/2500 970/1400 high pressure common rail

3.1. Mathematical Model Validation

As shown in Figure 5 (in this paper, A indicates the values after correction, B indicates the values before correction, T indicates the values obtained from equipment testing, C indicates the values obtained from calculation and M indicates the values obtained from the maps). The UWS injection was started at the 6th second. It can be seen in Figure 5 that the CN,Act curve declined gradually in the first 30 s, became stable at a very low level in the second 30 s, and two noticeable humps can be observed in the last 60 s. Considering the NH3 cross sensitivity of the NOx sensor, it could be initially assumed that the conversion efficiency was low in the first 30 s as the UWS injection was not sufficient. The NH3 slip increased significantly in the position of the two humps with the severely overloaded UWS injection. The UWS correction under dynamic conditions is critical for improving SCR conversion efficiency and NH3 slip inhibition.

Figure 5. NOx emissions in dynamic conditions before the correction.

The change rules of the four kinds of conversion efficiency which have been discussed in Section 2.1 are shown in Figure 6.

(1) In the beginning, PCon,Abs was less than the target valve PCon,Trg. However, these two values are the same as that after the 30th second.

Figure 5. NOx emissions in dynamic conditions before the correction.

The change rules of the four kinds of conversion efficiency which have been discussed inSection 2.1 are shown in Figure 6.

(1) In the beginning, PCon,Abs was less than the target valve PCon,Trg. However, these two values arethe same as that after the 30th second.

(2) PCon,Fuz and PCon,Abs remain the same as that before the 60th second. Then, two serious sinksappeared in the PCon,Fuz curve.

(3) After the beginning of UWS injection, PCon,Rel was stable near a 0 value between the 20th and60th second and fluctuated in a range of ±30% between the 60th and 120th second.

Energies 2017, 10, 12 11 of 17

Energies 2017, 10, 12 11 of 17

(2) PCon,Fuz and PCon,Abs remain the same as that before the 60th second. Then, two serious sinks appeared in the PCon,Fuz curve.

(3) After the beginning of UWS injection, PCon,Rel was stable near a 0 value between the 20th and 60th second and fluctuated in a range of ±30% between the 60th and 120th second.

Figure 6. The conversion efficiency tangents.

As a result, between the 30th and 60th second, the NH3 slip is assumed to be zero thus the UWS needs no correction. Between 60th and 90th or 100th and 120th second, NH3 slip is severely increased, thus UWS injection should be reduced. Between the 0th and 30th second there is no NH3 slip, thus UWS compensation be continued. The calculated value and actual experimental value of the NOx and NH3 are compared in the low-frequency process. The results are shown in Figure 7.

(1) From the 0 to the 60th second and the 90th to 100th second, the experimental value of the NH3 concentration downstream from the SCR system is almost 0 ppm. The NOx concentration and NH3 concentration calculated by the correction model completely overlap with the experimental values.

(2) From the 60th to 90th second and the 100th to 120th second, there are slight deviations in the hump position between the calculated value and the experimental value of the NH3 concentration. The two compared values of NOx concentration no longer completely overlap, but the range and the change rate are apparently the same.

(3) At zero NH3 slip condition, the calculation deviation of NH3 concentration was between −10 and 0 ppm and that of the NOx concentration was between −20 and 20 ppm.

(4) Under high NH3 slip conditions, the calculation deviation of NH3 concentration was between −40 and 100 ppm and that of NOx concentration was between −70 and 70 ppm.

Figure 7. The values and deviations of the NOx and NH3 before the correction.

Figure 6. The conversion efficiency tangents.

As a result, between the 30th and 60th second, the NH3 slip is assumed to be zero thus the UWSneeds no correction. Between 60th and 90th or 100th and 120th second, NH3 slip is severely increased,thus UWS injection should be reduced. Between the 0th and 30th second there is no NH3 slip, thusUWS compensation be continued. The calculated value and actual experimental value of the NOx andNH3 are compared in the low-frequency process. The results are shown in Figure 7.

(1) From the 0 to the 60th second and the 90th to 100th second, the experimental value of the NH3

concentration downstream from the SCR system is almost 0 ppm. The NOx concentrationand NH3 concentration calculated by the correction model completely overlap with theexperimental values.

(2) From the 60th to 90th second and the 100th to 120th second, there are slight deviations in thehump position between the calculated value and the experimental value of the NH3 concentration.The two compared values of NOx concentration no longer completely overlap, but the range andthe change rate are apparently the same.

(3) At zero NH3 slip condition, the calculation deviation of NH3 concentration was between −10and 0 ppm and that of the NOx concentration was between −20 and 20 ppm.

(4) Under high NH3 slip conditions, the calculation deviation of NH3 concentration was between−40 and 100 ppm and that of NOx concentration was between −70 and 70 ppm.

Energies 2017, 10, 12 11 of 17

(2) PCon,Fuz and PCon,Abs remain the same as that before the 60th second. Then, two serious sinks appeared in the PCon,Fuz curve.

(3) After the beginning of UWS injection, PCon,Rel was stable near a 0 value between the 20th and 60th second and fluctuated in a range of ±30% between the 60th and 120th second.

Figure 6. The conversion efficiency tangents.

As a result, between the 30th and 60th second, the NH3 slip is assumed to be zero thus the UWS needs no correction. Between 60th and 90th or 100th and 120th second, NH3 slip is severely increased, thus UWS injection should be reduced. Between the 0th and 30th second there is no NH3 slip, thus UWS compensation be continued. The calculated value and actual experimental value of the NOx and NH3 are compared in the low-frequency process. The results are shown in Figure 7.

(1) From the 0 to the 60th second and the 90th to 100th second, the experimental value of the NH3 concentration downstream from the SCR system is almost 0 ppm. The NOx concentration and NH3 concentration calculated by the correction model completely overlap with the experimental values.

(2) From the 60th to 90th second and the 100th to 120th second, there are slight deviations in the hump position between the calculated value and the experimental value of the NH3 concentration. The two compared values of NOx concentration no longer completely overlap, but the range and the change rate are apparently the same.

(3) At zero NH3 slip condition, the calculation deviation of NH3 concentration was between −10 and 0 ppm and that of the NOx concentration was between −20 and 20 ppm.

(4) Under high NH3 slip conditions, the calculation deviation of NH3 concentration was between −40 and 100 ppm and that of NOx concentration was between −70 and 70 ppm.

Figure 7. The values and deviations of the NOx and NH3 before the correction. Figure 7. The values and deviations of the NOx and NH3 before the correction.

Energies 2017, 10, 12 12 of 17

3.2. Low-Frequency Dynamic Process Correction Result

The low-frequency process has been repeated with the dynamic correction of the UWS injection.The NOx and NH3 emissions after the correction are shown in Figure 8.

Energies 2017, 10, 12 12 of 17

3.2. Low-Frequency Dynamic Process Correction Result

The low-frequency process has been repeated with the dynamic correction of the UWS injection. The NOx and NH3 emissions after the correction are shown in Figure 8.

Figure 8. The NOx and NH3 emissions before and after the correction.

(1) From the UWS injection starting position to 60th second there was no NH3 slip. Between the 60th and 120th second the second two humps appeared in the NH3 concentration curve at less than 80 ppm. That’s because of the fractionally unreacted NH3 slip downstream of the SCR system.

(2) In the low-frequency process, the actual tested value of the NOx emission was almost the same as the target value. The actual tested value was slightly lower than the target value in the hump region of the NH3 slip. That’s because of the undue UWS injection and more NOx was restored by the excessive NH3.

(3) In the hump region of the NH3 slip, the measured value is slightly higher than the target value of the NOx concentration. That’s because the NOx sensor was influenced by NH3 and NOx at the same time as shown in Equation (7).

Comparing the engine emissions before and after the UWS dynamic correction:

(1) The CNOx,Act was slightly reduced in the last 60 s on application of the UWS dynamic correction. However, the control method has no significant effect on the value of CNOx,Act in the low-frequency process.

(2) The CNH3,Act was also significantly reduced in the last 60 s after the UWS dynamic correction application. The values of CNH3,Act in the low-frequency process are also greatly influenced by the application of the correction.

(3) Overall, the application of UWS dynamic control method has reduced ∫CNH3,Act dt by 92.68%, respectively. Moreover it has also improved ∫CNOx,Act dt by 5.58%.

The change trends of the all four kinds of conversion efficiencies are compared after applying the control method as shown in Figure 9.

Figure 9. The four kinds of conversion efficiency after the correction.

Figure 8. The NOx and NH3 emissions before and after the correction.

(1) From the UWS injection starting position to 60th second there was no NH3 slip. Between the60th and 120th second the second two humps appeared in the NH3 concentration curve atless than 80 ppm. That’s because of the fractionally unreacted NH3 slip downstream of theSCR system.

(2) In the low-frequency process, the actual tested value of the NOx emission was almost the same asthe target value. The actual tested value was slightly lower than the target value in the humpregion of the NH3 slip. That’s because of the undue UWS injection and more NOx was restoredby the excessive NH3.

(3) In the hump region of the NH3 slip, the measured value is slightly higher than the target value ofthe NOx concentration. That’s because the NOx sensor was influenced by NH3 and NOx at thesame time as shown in Equation (7).

Comparing the engine emissions before and after the UWS dynamic correction:

(1) The CNOx,Act was slightly reduced in the last 60 s on application of the UWS dynamiccorrection. However, the control method has no significant effect on the value of CNOx,Act

in the low-frequency process.(2) The CNH3,Act was also significantly reduced in the last 60 s after the UWS dynamic correction

application. The values of CNH3,Act in the low-frequency process are also greatly influenced bythe application of the correction.

(3) Overall, the application of UWS dynamic control method has reduced∫

CNH3,Act dt by 92.68%,respectively. Moreover it has also improved

∫CNOx,Act dt by 5.58%.

The change trends of the all four kinds of conversion efficiencies are compared after applying thecontrol method as shown in Figure 9.

(1) In the beginning of the low-frequency process, PCon,Abs was smaller than PCon,Trg. However, thetwo curves overlapped after the 15th second.

(2) From 0 to 60th second, PCon,Fuz was the same as PCon,Abs, whereas, after the 60th second PCon,Fuz

started to deviate with a slightly sinking trend.(3) Initially PCon,Abs was less than 0 with a rising trend, whereas, it became stable when close to

0 and 15th to 60th second, while from 60th to 120th second it fluctuated many times with anamplitude between −15% and 15%.

Energies 2017, 10, 12 13 of 17

(4) PCon,Abs and PCon,Trg were achieved in a shorter period as compared to Figure 4. The sinkingamplitude of the PCon,Fuz curve has significantly decreased after the 60th second and becamestable in the last 60 s.

Energies 2017, 10, 12 12 of 17

3.2. Low-Frequency Dynamic Process Correction Result

The low-frequency process has been repeated with the dynamic correction of the UWS injection. The NOx and NH3 emissions after the correction are shown in Figure 8.

Figure 8. The NOx and NH3 emissions before and after the correction.

(1) From the UWS injection starting position to 60th second there was no NH3 slip. Between the 60th and 120th second the second two humps appeared in the NH3 concentration curve at less than 80 ppm. That’s because of the fractionally unreacted NH3 slip downstream of the SCR system.

(2) In the low-frequency process, the actual tested value of the NOx emission was almost the same as the target value. The actual tested value was slightly lower than the target value in the hump region of the NH3 slip. That’s because of the undue UWS injection and more NOx was restored by the excessive NH3.

(3) In the hump region of the NH3 slip, the measured value is slightly higher than the target value of the NOx concentration. That’s because the NOx sensor was influenced by NH3 and NOx at the same time as shown in Equation (7).

Comparing the engine emissions before and after the UWS dynamic correction:

(1) The CNOx,Act was slightly reduced in the last 60 s on application of the UWS dynamic correction. However, the control method has no significant effect on the value of CNOx,Act in the low-frequency process.

(2) The CNH3,Act was also significantly reduced in the last 60 s after the UWS dynamic correction application. The values of CNH3,Act in the low-frequency process are also greatly influenced by the application of the correction.

(3) Overall, the application of UWS dynamic control method has reduced ∫CNH3,Act dt by 92.68%, respectively. Moreover it has also improved ∫CNOx,Act dt by 5.58%.

The change trends of the all four kinds of conversion efficiencies are compared after applying the control method as shown in Figure 9.

Figure 9. The four kinds of conversion efficiency after the correction. Figure 9. The four kinds of conversion efficiency after the correction.

The calculated values and experimental values of the NOx and NH3 were further compared toensure that the revised data can be well trusted as shown in Figure 10.

Energies 2017, 10, 12 13 of 17

(1) In the beginning of the low-frequency process, PCon,Abs was smaller than PCon,Trg. However, the two curves overlapped after the 15th second.

(2) From 0 to 60th second, PCon,Fuz was the same as PCon,Abs, whereas, after the 60th second PCon,Fuz started to deviate with a slightly sinking trend.

(3) Initially PCon,Abs was less than 0 with a rising trend, whereas, it became stable when close to 0 and 15th to 60th second, while from 60th to 120th second it fluctuated many times with an amplitude between −15% and 15%.

(4) PCon,Abs and PCon,Trg were achieved in a shorter period as compared to Figure 4. The sinking amplitude of the PCon,Fuz curve has significantly decreased after the 60th second and became stable in the last 60 s.

The calculated values and experimental values of the NOx and NH3 were further compared to ensure that the revised data can be well trusted as shown in Figure 10.

Figure 10. The values and deviations of the NOx and NH3 after the correction.

(1) In the corrected low-frequency process the trend of the calculated NH3 concentration was the same with that of the experimental value. The deviation of the calculation was more obvious in the region of high NH3 slip as compared to the results shown in Figure 7. The deviation oscillated between −10 to 0 ppm and −5 to 0 ppm with NH3 slip and between −40 to 100 ppm and −20 to 15 ppm without NH3 slip.

(2) Moreover, during the corrected low−frequency process trend of the calculated NOx concentration was the same as the actual tested value. The calculation deviation was uniformly distributed and oscillated between −70 to 70 ppm and −20 to 20 ppm as compared with Figure 7.

(3) NOx concentration calculation and NH3 concentration downstream of the SCR system would become more accurate with application of the UWS dynamic correction.

It is compared by the correction factor φ before and after applying the UWS dynamic control method as illustrated in Figure 11 (φA and φB indicate the correction factors after and before applying the control method, respectively).

(1) From the UWS injection starting to 20th second, φB remained at more than 0 with a gradually declining trend. That is for the catalyst NH3 storage characteristic therefore UWS injection should be compensated.

(2) From 20th to 60th second, φB remained constant and close to 0. Now catalyst NH3 storage has been saturated without NH3 slip so UWS injection may not be corrected.

(3) From 60th to 120th second, φB was less than 0. As compared to Figure 7 it is observed that the change trend of φB was contrary to the change trend of NH3 concentration. It is because of the severely increasing NH3 slip and UWS injection must be reduced.

(4) φA and φB were greater than 0 and declined gradually from the starting position of UWS injection till the 20th second. However, the decrease of φA was faster than that of φB.

(5) φA and φB remained close to 0 from 20th till the 60th second.

Figure 10. The values and deviations of the NOx and NH3 after the correction.

(1) In the corrected low-frequency process the trend of the calculated NH3 concentration was thesame with that of the experimental value. The deviation of the calculation was more obvious inthe region of high NH3 slip as compared to the results shown in Figure 7. The deviation oscillatedbetween −10 to 0 ppm and −5 to 0 ppm with NH3 slip and between −40 to 100 ppm and −20 to15 ppm without NH3 slip.

(2) Moreover, during the corrected low−frequency process trend of the calculated NOx concentrationwas the same as the actual tested value. The calculation deviation was uniformly distributed andoscillated between −70 to 70 ppm and −20 to 20 ppm as compared with Figure 7.

(3) NOx concentration calculation and NH3 concentration downstream of the SCR system wouldbecome more accurate with application of the UWS dynamic correction.

It is compared by the correction factor ϕ before and after applying the UWS dynamic controlmethod as illustrated in Figure 11 (ϕA and ϕB indicate the correction factors after and before applyingthe control method, respectively).

(1) From the UWS injection starting to 20th second, ϕB remained at more than 0 with a graduallydeclining trend. That is for the catalyst NH3 storage characteristic therefore UWS injection shouldbe compensated.

Energies 2017, 10, 12 14 of 17

(2) From 20th to 60th second, ϕB remained constant and close to 0. Now catalyst NH3 storage hasbeen saturated without NH3 slip so UWS injection may not be corrected.

(3) From 60th to 120th second, ϕB was less than 0. As compared to Figure 7 it is observed that thechange trend of ϕB was contrary to the change trend of NH3 concentration. It is because of theseverely increasing NH3 slip and UWS injection must be reduced.

(4) ϕA andϕB were greater than 0 and declined gradually from the starting position of UWS injectiontill the 20th second. However, the decrease of ϕA was faster than that of ϕB.

(5) ϕA and ϕB remained close to 0 from 20th till the 60th second.(6) Values of both the factors (ϕA and ϕB) became less than 0 during the last 60 s. Both factors shared

two troughs. The trough values of ϕB ranged from −5 to −8 and that of the ϕA was −1 to 0in curve.

Energies 2017, 10, 12 14 of 17

(6) Values of both the factors (φA and φB) became less than 0 during the last 60 s. Both factors shared two troughs. The trough values of φB ranged from −5 to −8 and that of the φA was −1 to 0 in curve.

Figure 11. The dynamic correction fact before and after the correction.

The uncertain conversion efficiency PCon,Fuz and its velocity vCon,Fuz before and after applying the UWS dynamic correction are compared as shown in Figure 12:

(1) PCon,Fuz and vCon,Fuz were changed in the last 60 s on applying UWS injection dynamic correction during a high level of NH3 slip.

(2) The PCon,Fuz after correction appeared more close to PCon,Trg and its range was reduced from −250%–95% to 40%–95% compared to the values before correction.

(3) The range of the vCon,Fuz was reduced from −70%–30% to −10%–10% compared to the value before correction.

Figure 12. Conversion efficiency and its velocity before and after correction.

3.3. High-Frequency Dynamic Process Correction Result

The emission data comparison in ETC cycle before and after the UWS dynamic control method application is shown in Figure 13.

Figure 11. The dynamic correction fact before and after the correction.

The uncertain conversion efficiency PCon,Fuz and its velocity vCon,Fuz before and after applyingthe UWS dynamic correction are compared as shown in Figure 12:

(1) PCon,Fuz and vCon,Fuz were changed in the last 60 s on applying UWS injection dynamic correctionduring a high level of NH3 slip.

(2) The PCon,Fuz after correction appeared more close to PCon,Trg and its range was reduced from−250%–95% to 40%–95% compared to the values before correction.

(3) The range of the vCon,Fuz was reduced from −70%–30% to −10%–10% compared to the valuebefore correction.

Energies 2017, 10, 12 14 of 17

(6) Values of both the factors (φA and φB) became less than 0 during the last 60 s. Both factors shared two troughs. The trough values of φB ranged from −5 to −8 and that of the φA was −1 to 0 in curve.

Figure 11. The dynamic correction fact before and after the correction.

The uncertain conversion efficiency PCon,Fuz and its velocity vCon,Fuz before and after applying the UWS dynamic correction are compared as shown in Figure 12:

(1) PCon,Fuz and vCon,Fuz were changed in the last 60 s on applying UWS injection dynamic correction during a high level of NH3 slip.

(2) The PCon,Fuz after correction appeared more close to PCon,Trg and its range was reduced from −250%–95% to 40%–95% compared to the values before correction.

(3) The range of the vCon,Fuz was reduced from −70%–30% to −10%–10% compared to the value before correction.

Figure 12. Conversion efficiency and its velocity before and after correction.

3.3. High-Frequency Dynamic Process Correction Result

The emission data comparison in ETC cycle before and after the UWS dynamic control method application is shown in Figure 13.

Figure 12. Conversion efficiency and its velocity before and after correction.

Energies 2017, 10, 12 15 of 17

3.3. High-Frequency Dynamic Process Correction Result

The emission data comparison in ETC cycle before and after the UWS dynamic control methodapplication is shown in Figure 13.Energies 2017, 10, 12 15 of 17

(a) (b)

Figure 13. Emissions in the engine ETC cycle. (a) Before UWS injection dynamic correction; and (b) after UWS injection dynamic correction.

The result indicated a great high efficiency of NOx conversion in the two tests. The NOx emission was improved by 53.9% and the NH3 slip was reduced by 90.8% with the application of the control method. The NH3 slip was also inhibited significantly. The levels of engine NOx emissions and NH3 slip were improved and conform to the Euro-V standard.

4. Conclusions

(1) It can be generalized that the “uncertain conversion efficiency curve tangent analysis” method can accurately justify the different NH3 slip situation.

(2) The calculation deviation can be controlled with NOx between −20 ppm and 20 ppm and NH3 between −20 ppm to 15 ppm by application of UWS dynamic correction in a low-frequency process. The NOx emission was improved by 5.58% and NH3 slip was reduced by 92.68%.

(3) In the application of UWS dynamic correction in high-frequency process (i.e., during of ETC test), in spite the fact the NOx emission has been improved by 53.9%, the NH3 slip was reduced by 90.8%. The level of engine NOx emissions and NH3 slip has been improved up to Euro-IV and closer to the Euro-V standard. The newly developed method presents a significant NH3 slip inhibition.

Author Contributions: Long Li and Wei Lin made the method. Long Li and Wei Lin designed the experiment and organized the entire experiment process. Youtong Zhang made many experimental suggestions and collated the experiment data.

Conflicts of Interest: The authors declare no conflict of interests.

Abbreviations

ASC Ammonia slip catalysts ETC European transient cycle HCR High-pressure common rail PM Particulate matter SCR Select catalyst reduction UWS Urea water solution

Symbols

aUWS,Act Acceleration of qUWS,Act CN,Act NOx concentration measured by the NOx sensor CNOx,Act Actual NOx concentration at the testing position CNOx,Ori Original NOx concentration of the engine before aftertreatment CNOx,Trg The target of NOx concentration downstream SCR system CNH3,Act Actual NH3 concentration at the testing position CDNH3,Act Test error of actual NH3 concentration

CDNOx,Act Test error of actual NOx concentration KT Cross sensitive factor

PCon,Fuz Uncertain conversion efficiency

Figure 13. Emissions in the engine ETC cycle. (a) Before UWS injection dynamic correction; and (b)after UWS injection dynamic correction.

The result indicated a great high efficiency of NOx conversion in the two tests. The NOx emissionwas improved by 53.9% and the NH3 slip was reduced by 90.8% with the application of the controlmethod. The NH3 slip was also inhibited significantly. The levels of engine NOx emissions and NH3

slip were improved and conform to the Euro-V standard.

4. Conclusions

(1) It can be generalized that the “uncertain conversion efficiency curve tangent analysis” methodcan accurately justify the different NH3 slip situation.

(2) The calculation deviation can be controlled with NOx between −20 ppm and 20 ppm and NH3

between −20 ppm to 15 ppm by application of UWS dynamic correction in a low-frequencyprocess. The NOx emission was improved by 5.58% and NH3 slip was reduced by 92.68%.

(3) In the application of UWS dynamic correction in high-frequency process (i.e., during of ETCtest), in spite the fact the NOx emission has been improved by 53.9%, the NH3 slip was reducedby 90.8%. The level of engine NOx emissions and NH3 slip has been improved up to Euro-IVand closer to the Euro-V standard. The newly developed method presents a significant NH3

slip inhibition.

Author Contributions: Long Li and Wei Lin made the method. Long Li and Wei Lin designed the experimentand organized the entire experiment process. Youtong Zhang made many experimental suggestions and collatedthe experiment data.

Conflicts of Interest: The authors declare no conflict of interests.

Abbreviations

ASC Ammonia slip catalystsETC European transient cycleHCR High-pressure common railPM Particulate matterSCR Select catalyst reductionUWS Urea water solution

Energies 2017, 10, 12 16 of 17

Symbols

aUWS,Act Acceleration of qUWS,ActCN,Act NOx concentration measured by the NOx sensorCNOx,Act Actual NOx concentration at the testing positionCNOx,Ori Original NOx concentration of the engine before aftertreatmentCNOx,Trg The target of NOx concentration downstream SCR systemCNH3,Act Actual NH3 concentration at the testing positionCDNH3,Act Test error of actual NH3 concentrationCDNOx,Act Test error of actual NOx concentrationKT Cross sensitive factorPCon,Fuz Uncertain conversion efficiencyPCon,Abs Absolute conversion efficiencyPCon,Rel Relative conversion efficiencyPCon,Trg Targeted conversion efficiencyqUWS,Act Real-time UWS injection rate after correctionqUWS,Bas Basic UWS injection rate before correctionQNOx,Red NOx conversion potentialQNOx,AcRed Actual value of the total reduced NOxQNOx,TrgRed Total reduced NOx under target conversion efficiencyRAN Ammonia nitrogen ratio set in the SCR control strategyϕ Correction factor of the UWS injection rateϕA Correct factor after applying the control methodϕB Correct factor before applying the control method Reference

References

1. Johnsont, T.V. Diesel emission control in review. SAE Int. J. Fuels Lubr. 2009, 2. [CrossRef]2. Johnsont, T.V. Diesel emissions in review. SAE Int. J. Eng. 2011, 4, 143–157. [CrossRef]3. Koebel, M.; Elsener, M.; Kleemann, M. Urea-SCR: A promising technique to reduce NOx emissions from

automotive diesel engines. Catal. Today 2000, 59, 335–345. [CrossRef]4. Flynn, P.F.; Hunter, G.L.; Durrett, R.P.; Farrell, L.A.; Akinemi, W.C. Minimum engine flame temperature

impacts on diesel and spark-ignition engine NOx production. SAE Tech. Pap. 2000. [CrossRef]5. Sharp, C.; Howell, S.; Jobe, J. The effect of biodiesel fuels on transient emissions from modern diesel engines,

Part I: Regulated emissions and performance. SAE Tech. Pap. 2000. [CrossRef]6. Cloudt, R.; Baert, R.; Willems, F.; Vergouwe, M. SCR-only concept for heavy-duty Euro VI applications. MTZ

2009, 70, 58–63. [CrossRef]7. Seher, D.; Reichelt, M.; Wickert, S. Control strategy for NOx—Emission reduction with SCR. SAE Tech. Pap.

2003. [CrossRef]8. Kammerstetter, H.; Werner, M.; Doell, R.; Kanters, G. The challenge of precise characterizing the Specific

large-span flows in urea dosing systems for NOx reduction. SAE Tech. Pap. 2008. [CrossRef]9. Kass, M.D.; Thomas, J.F.; Lewis, S.A.; Storey, J.M.; Domingo, N.; Graves, R.L.; Panov, A.; Park, P. Selective

catalytic reduction of NOx emissions from a 5.9 liter diesel engine using ethanol as a reductant. SAE Tech. Pap.2003. [CrossRef]

10. Chen, C.-T.; Tan, W.-L. Mathematical modeling, optimal design and control of an SCR reactor for NOx

removal. J. Taiwan. Inst. Chem. Eng. 2012, 43, 409–419. [CrossRef]11. Nova, I.; Tronconi, E. (Eds.) Urea-SCR Technology for de NOx after Treatment of Diesel Exhausts; Spring: New

York, NY, USA, 2014.12. Shrestha, S.; Harold, M.P.; Kamasamudram, K. Experimental and modeling study of selective ammonia

oxidation on multi-functional washcoated monolith catalysts. Chem. Eng. J. 2015, 278, 24–35. [CrossRef]13. Opitz, B.; Bendrich, M.; Drochner, A.; Vogel, H.; Hayes, R.E.; Forbes, J.F.; Votsmeier, M. Simulation study

of SCR catalysts with individually adjusted ammonia dosing strategies. Chem. Eng. J. 2015, 264, 936–944.[CrossRef]

14. Feng, T.; Lü, L. The characteristics of ammonia storage and the development of model-based control fordiesel engine urea-SCR system. J. Ind. Eng. Chem. 2015, 28, 97–109. [CrossRef]

15. Rauch, D.; Albrecht, G.; Kubinski, D.; Moos, R. A microwave-based method to monitor the ammonia loadingfor a vanadia-based SCR catalyst. Appl. Catal. B 2015, 165, 36–42. [CrossRef]

Energies 2017, 10, 12 17 of 17

16. Zhang, H.; Wang, J. Ammonia coverage ratio and input simultaneous estimation in ground vehicle selectivecatalytic reduction (SCR) systems. J. Frankl. Inst. 2015, 352, 708–723. [CrossRef]

17. Schär, C.M.; Onder, C.H.; Geering, H.P.; Elsener, M. Control-oriented model of an SCR catalytic convertersystem. SAE Tech. Pap. 2004. [CrossRef]

18. Willems, F.; Cloudt, R.; van den Eijinden, E.; van Genderen, M.; Verbeek, R.; de Jager, B.; Boomsma, W.;van de Heuvel, I. Is closed-loop SCR control required to meet future emission targets? SAE Tech. Pap. 2007.[CrossRef]

19. Wang, D.Y.; Yao, S.; Shost, M.; Yoo, J.-H.; Cabush, D.; Racine, D.; Cloudt, R.; Willems, F. Ammonia sensor forclose-loop SCR control. SAE Tech. Pap. 2008. [CrossRef]

20. Kobayashi, N.; Yamashita, A.; Naito, O.; Setoguchi, T.; Murase, T. Development of simultaneous NOx/NH3

sensor in exhaust gas. Mitsubishi Tech. Rev. 2001, 38, 126–130.21. Moos, R. A brief overview on automotive exhaust gas sensors based on electroceramics. Int. J. Appl.

Ceram. Technol. 2005, 2, 401–413. [CrossRef]22. Lu, G.; Diao, Q.; Yin, C.G.; Yang, S.Q.; Guan, Y.Z.; Cheng, X.Y.; Liang, X.S. High performance mixed-potential

type NOx sensor based on stabilized zironia and oxide electrode. Solid State Ion. 2014, 262, 292–297.[CrossRef]

23. Jeffrey, W.F. Materials for high temperature electrochemical NOx gas sensors. Sens. Actuators B 2007, 121,652–663.

24. Wang, W. Simulation and Experimental Research on SCR System Control Strategy of Marine Diesel Engine.Master’s Thesis, Harbin Engineering University, Harbin, China, 2013.

© 2016 by the authors; licensee MDPI, Basel, Switzerland. This article is an open accessarticle distributed under the terms and conditions of the Creative Commons Attribution(CC-BY) license (http://creativecommons.org/licenses/by/4.0/).